LB: How the measurements were derived

LB: How the measurements were derived

This is an explanation (using trial 1, before cooking as an example) of various techniques used to get the results. Some are self explanatory/simple enough to cover in the list of steps, and these are the rest, updated as experiments progress. Note: even though the example is trial 1, image-j was actually used until after trial 3.

1. Surface area- manual and image-j
2. Making of flats from trial 3 onwards
3. Weighing with spring balance
4. Correlation formula/Graphs

1. Surface area

I have used two ways, manually and with image-j software, to find the surface area of each GRB, though how successful each is has yet to be evaluated. In the list of steps, such a photograph was to be taken:

fom trial 1, before cooking, taken with camera held parallel from ground

Manual

Using powerpoint, a ruler and colour-coded squares encircling each GRB was superimposed onto the photograph. 

After the photo has been marked

The ruler from the original photo was to indicate the scale of the photo. The lines of the square extended until the ruler, where the diameter of each GRB is read off two times, length and width wise. Taking the average, surface area is calculated with the formulas πr² (circle) for the flats, and 4πr² (sphere) for the balls. 
Figure 1: Manual calculations for surface area before cooking, trial 1

	Radius before cooking/cm (rounded to 1 dp)
Type of GRB	Derived from length	Derived from width	Average	Surface area before cooking/ cm2 (rounded to 1 dp)
Green 40g flat	3	3.1	3.1	120.8
White 40g flat	2.7	2.9	2.8	98.5
Green 40g ball	2.4	2.5	2.5	78.5
White 40g ball	2	2.2	2.1	55.4
Green 20g ball	1.8	2.1	2	50.3
White 20g ball	1.6	1.6	1.6	32.2

Image-J

Image-J is a free, downloadable software capable of analysing images. This is the developer's website. After the photograph was opened in image-J, I used the set-scale function to declare the number of pixels in 1 cm by drawing to scale a 1 cm line on the ruler. Then, I used the free-hand drawing tool to trace the shape of each GRB and the measure function to analyse and convert into cm² how many pixels the GRB was.

Figure 2: Image-J calculations for surface area before cooking, trial 1

	Surface Area/ cm2 calculated by image-j
Type of GRB	Before cooking
Green 40g flat	32.205
White 40g flat	25.496
Green 40g ball	18.948
White 40g ball	13.867
Green 20g ball	12.027
White 20g ball	9.579

Conclusion

• Because it calculates 2D area, image-J is not applicable for the spherical GRB and there is a great difference between image-J values and manual values. So, I've decided to forgo image-j for future trials and only use the manual method.
• The "spheres" were not perfect spheres. Because of the pliability of the douch, they had no structural integrity and easily lost their shape. Hence, no more tests using round GRB after trial 1.  

2. Making of flats

After trial 1 and 2 measurements were analysed, I realised that using hands to shape the flats led to inconsistency in the shape. After talking with a friend (Zhi Rui), we came up with a way to improve that.

Diagram of the process

Before placing the spherical GRB in, the set-up is covered in cling wrap to prevent sticking.

The blocks are anything (e.g. bowls, cups, pieces of wood), but each set is identical in height and the plate has a flat bottom. This method will make the force applied more even.

3. Weighing with spring balance

Photo from real life

The GRB is placed in a plastic “pouch”, then the hook of the spring balance is forced through the top, leaving the pouch hanging. The measurement is then read off the spring balance and the mass of the plastic is assumed to be negligible.


4. Correlation formula/Graphs

Correlation was calculated with the =CORREL(array1,array2) function in excel, which uses  (pearson product-moment correlation coefficient) as the formula. This has been chosen because the data to be analysed mostly fits Pearson's assumptions that it has interval/ratio measurements, a linear relationship, minimal outliers and homoscedasticity (consistent variance along the trend line). 

example

The graphs for the final experiment are created using Apache OpenOffice instead of the previous Excel.